build failure on solaris-i586 in make/sun/cldr
naoto.sato at oracle.com
Wed Dec 12 19:49:34 UTC 2012
Looks like the "$(CD) $$dir;" is unnecessary here. Thanks for the catch.
Just wondering why it is failing on solaris-i586 only. I don't use
ALT_OUTPUTDIR, but never seen this failure on my environment.
Can you please file a bug for this?
On 12/11/12 11:33 PM, Weijun Wang wrote:
> I haven't build on solaris-i586 for some time and see a failure today in
> make/sun/cldr. The Makefile  has these lines:
> 75 for dir in $(GENSRCDIR); do \
> 76 if [ -d $$dir ] ; then \
> 77 ( $(CD) $$dir; \
> 78 for sdir in $(CLDRGENSRCDIR); do \
> 79 if [ -d $$sdir ] ; then \
> 80 $(FIND) $$sdir \
> 81 -name '*.java' -print >>
> $(JAVA_SOURCE_LIST) ; \
> 82 fi ; \
> 83 done \
> 84 ); \
> 85 fi; \
> 86 done \
> So it goes into $(GENSRCDIR) and then tries to look for files inside
> (one of) $(CLDRGENSRCDIR). The latter is defined as
> 49 CLDRGENSRCDIR = $(GENSRCDIR)/sun/text/resources/cldr \
> 50 $(GENSRCDIR)/sun/util/cldr \
> 51 $(GENSRCDIR)/sun/util/resources/cldr
> in the same file.
> In my build, GENSRCDIR is something like
> ../../../build/solaris-i586/gensrc. Since this is a relative directory,
> you cannot cd into it and use it again.
> Maybe the first CD is just useless.
> Is everyone using ALT_OUTPUTDIR?
More information about the core-libs-dev